mirror of
https://github.com/deepseek-ai/FlashMLA
synced 2025-06-26 18:15:54 +00:00
use fa'3 transv
This commit is contained in:
@@ -33,7 +33,7 @@ def cal_diff(x: torch.Tensor, y: torch.Tensor, name: str) -> None:
|
||||
cos_diff = 1 - 2 * (x * y).sum().item() / max((x * x + y * y).sum().item(), 1e-12)
|
||||
amax_diff = (x - y).abs().max().item()
|
||||
# print(f"{name}: {cos_diff=}, {RMSE=}, {amax_diff=}")
|
||||
#assert cos_diff < 1e-5
|
||||
assert cos_diff < 1e-5
|
||||
|
||||
|
||||
@torch.inference_mode()
|
||||
|
||||
Reference in New Issue
Block a user